An approach for segmentation of medical images using pillar K-means algorithm
International Journal of Computer Trends and Technology (IJCTT) - volume4Issue4 –April 2013
An approach for segmentation of medical images using pillar K-means algorithm
M.Pavani#1, Prof. S.Balaji*2
#
Department of Electronics and Computers Engineering, K.L. University
Vaddeswaram, Vijayawada, India.
*
Department of Electronics and Computers Engineering, K.L. University
Vaddeswaram, Vijayawada, India.
ABSTRACT
This paper presents an approach for image segmentation using pillar K-Means algorithm. In this paper the segmentation process includes a mechanism for clustering the elements of high resolution images. By using this process we can improve precision and reduce computational time. The system applies K-means clustering to image segmentation after optimized by pillar algorithm. The pillar algorithm considers that pillars placement should be located as far as possible from each other. The pillars placement is located far from each other to withstand against the pressure distribution of a roof, as identical to number of centroids among the data distribution. This algorithm is able to optimize the Kmeans clustering for image segmentation in terms of precision and computational time. By calculating the accumulated distance metric between each data point and all previous centroids it designates the initial centroids position and then it selects the data points which have maximum distance as new initial centroids. According to accumulated distance metric all the initial centroids are distributed in his algorithm. This paper evaluates by using an existing approach for image segmentation. But here we use medical images for segmentation. The experimental results clarify that this approach improves the segmentation quality in terms of precision and computational time.
Keywords - Image segmentation, K-means clustering,
Pillar algorithm.
I. INTRODUCTION
In computer vision, image segmentation[2] is the process of partitioning a
An approach for segmentation of medical images using pillar K-means algorithm
M.Pavani#1, Prof. S.Balaji*2
#
Department of Electronics and Computers Engineering, K.L. University
Vaddeswaram, Vijayawada, India.
*
Department of Electronics and Computers Engineering, K.L. University
Vaddeswaram, Vijayawada, India.
ABSTRACT
This paper presents an approach for image segmentation using pillar K-Means algorithm. In this paper the segmentation process includes a mechanism for clustering the elements of high resolution images. By using this process we can improve precision and reduce computational time. The system applies K-means clustering to image segmentation after optimized by pillar algorithm. The pillar algorithm considers that pillars placement should be located as far as possible from each other. The pillars placement is located far from each other to withstand against the pressure distribution of a roof, as identical to number of centroids among the data distribution. This algorithm is able to optimize the Kmeans clustering for image segmentation in terms of precision and computational time. By calculating the accumulated distance metric between each data point and all previous centroids it designates the initial centroids position and then it selects the data points which have maximum distance as new initial centroids. According to accumulated distance metric all the initial centroids are distributed in his algorithm. This paper evaluates by using an existing approach for image segmentation. But here we use medical images for segmentation. The experimental results clarify that this approach improves the segmentation quality in terms of precision and computational time.
Keywords - Image segmentation, K-means clustering,
Pillar algorithm.
I. INTRODUCTION
In computer vision, image segmentation[2] is the process of partitioning a
References: Recognition. Oxford, England: Oxford University Press, 1995, http://mathworld.wolfram.com/K-Means Clustering System with 3D Color Vector Quantization and Clusterbased Shape and Structure Features”, The 19th EuropeanJapanese Conference on Information Modelling and Knowledge Bases, Maribor, 2009.